Columbia University School of International and Public Affairs Capstone Project

Client: Moody’s Investor Services Inc.

Debt Intolerance in Latin America

By: Amir Safa, Seda Turksever, Juan Pereira, Eduardo Granizo, Nathan Fabius and Quentin Dumont

Faculty Advisor: Adjunct Professor Gray Newman

May 2, 2017

DISCLAIMER: This study represents the views of the authors and does not represent views of Moody’s Investors Service, Inc. (MIS) or Columbia University. The views expressed herein should be attributed to the authors and not to MIS or Columbia University. MIS provided feedback and observations to the student authors for educational purposes only but is not responsible or liable in any way for the content of the study. Should you wish to contact the authors, please email [email protected], [email protected], [email protected], [email protected], [email protected] and [email protected].

1

Table of Contents

Introduction 3 Literature Review 3 Data Analysis: Latin America 14 Case Study: 17 Case Study: Chile 20 Case Study: 23 Case Study: US 26 Appendix 31

2

Introduction Moody’s Latin America Desk asked the Capstone team to examine the presence of intolerance in Latin America and to research factors which may help explain its persistence. Some economists describe “debt intolerance” as the inability of economies to manage levels of overall debt that, under the similar circumstances, would be easily manageable for developed countries.1 For example, during the Great , US total public debt to GDP nearly doubled in a short period of time from 62 percent in 2007 to 100 percent in 2012.2 Yet, the US government did not experience severe economic stress. On the other hand, most - if not all - Latin American economies would not have been able to tolerate such high levels of debt to GDP. Mexico in 1995 with a much lower level of debt to GDP of 34 percent experienced extreme stress and required support from the US government as well as other international financial institutions including the International Monetary Fund.3 While analysts have researched debt intolerance, there is still much debate surrounding this phenomenon. Using the following methods, this report examines factors thought to be most relevant to debt intolerance: (1) providing a literature review on leading publications; (2) recreating a model that regresses credit ratings on a number of indicators of debt intolerance; (3) conducting a sensitivity analysis on Latin American countries on selected indicators of debt intolerance; (4) discussing factors in specific context using three Latin American cases: Mexico, Chile and Brazil, as well as the US to provide a contrasting example of an advanced economy that does not suffer from debt intolerance.

Literature Review This section of the report identifies and summarizes these pieces of literature on debt intolerance:

1) “Debt Intolerance” (Reinhart, Rogoff and Savastano 2003) further expanded in the book This Time is Different (Reinhart and Rogoff 2009), which together form the most seminal work on debt intolerance; 2) “A Debt Intolerance Framework Applied to Central America, Panama and the Dominican Republic” (Bannister and Barrot 2011) which provides the most detailed and advanced revisions of the aforementioned works from Reinhart et al.; 3) “The Causes of Sovereign Defaults: Ability to Manage Crises Not Merely Determined by Debt Levels” (Duggar et al. 2010), one of the most relevant publications on made available from Moody’s; and 4) “The Institutional Determinants of Debt Intolerance” (Giordano and Tommasino 2011) used in this report to improve existing models by including factors related to institutional strength. The literature review illustrates that there is no systematic approach to understand the mechanics of debt intolerance; however, much of the prominent work uses a “decidedly empirical”4 approach drawing observations from analyses of datasets. There is still work to be done to link up the 3

empirical work with a coherent theoretical framework.

(1) “Debt Intolerance” (Reinhart, Rogoff and Savastano 2003); and This Time is Different (Reinhart and Rogoff 2009) In 2003, leading economists Carmen Reinhart, Kenneth Rogoff and Miguel Savastano introduce and coin the term “debt intolerance.” In 2009, Reinhart and Rogoff followed up with the most comprehensive study ever conducted at that time on debt intolerance, culminating in the book This Time is Different. These works are considered the most prominent pieces of literature on this topic.

In the paper “Debt Intolerance” the authors argue that “history matters” when considering a country’s ability to take on moderate to high levels of domestic and .5 Specifically, they argue that “safe” external debt-to-GDP thresholds for debt intolerant countries are low and could be as low as 15 percent in some instances.6 The authors conclude that the data overwhelmingly suggests “the thresholds for emerging market economies with high debt intolerance are much lower than those for advanced industrial economies or for those emerging market economies that have never defaulted on their external debt.”7 The thresholds for these countries often depend on the country’s history of default and .8 A history of serial default is linked to a country’s level of vulnerability to a as its external mount.9 Furthermore, they note debt intolerant countries often experience institutional weaknesses in fiscal structures and financial systems.10 One of their key findings is that debt intolerance can be explained by a small number of factors associated with repayment history, indebtedness level, and history of macroeconomic stability.11

Reinhart, Rogoff and Savastano developed a model in an attempt to quantify debt intolerance for the first time.12 They examine over 100 countries and look specifically at the history of credit events with data going dating back to the 1820s. The model uses a regression analysis to examine the relationship between the Institutional Investor Rating (IIR) on external debt to GNP ratio, default and inflation history. Compiled by the magazine Institutional Investor from a number of global banks and securities firms, the IIR is a metric, which grades countries on a scale of 0 to 100 - with a high score signaling low default probability.13 In their analysis, they examine the extent to which the IIR changes with an additional unit of debt.14

Reinhart and Rogoff significantly expand this initial study in the book This Time is Different. The authors cover eight centuries in providing a quantitative history of financial crises and identifying similarities among them. They expand on their initial argument that “history matters” concluding in the book among the wide range of crises that excessive debt accumulation, whether it be by the government, banks, corporations, or consumers often poses greater systemic risks than it seems during a boom. Among the financial crises analyzed is sovereign debt default, which the authors believe arise for several reasons including: (1) lenders often cannot enforce debt contracts across national borders; (2) shifting political environments including during the time of elections; and (3) a slowdown in financial centers that hit emerging markets that rely heavily on exports and commodities.15

4

(2) A Debt Intolerance Framework Applied to Central America, Panama and the Dominican Republic (Bannister and Barrot 2011) The most important review and revision to the Reinhart, Rogoff and Savastano model can be found in the IMF 2011 paper “Debt Intolerance Framework Applied to Central America, Panama and the Dominican Republic” by Geoffrey J. Bannister and Luis-Diego Barrot. While the authors keep a similar approach of regressing credit ratings on a host of macroeconomic variables and debt repayment history, they have adjusted the Reinhart, Rogoff and Savastano model to overcome some of its shortcomings. Bannister and Barrot sought in their model to reduce endogeneity in debt, inflation and default; take into account changes in the IIR and debt over time; and include domestic debt in addition to external debt to provide a more complete picture of debt intolerance.16 Three of the most important modifications to the Reinhart, Rogoff and Savastano model are discussed here. First, they remove Reinhart, Rogoff and Savastano’s inclusion of IIR rating groups as a variable in the initial equation, which risked explaining IIR ratings by IIR ratings themselves. Second, the authors include time and country fixed effects in their regression. Country fixed effects control for unobserved national features that explain IIR rating but do not move across time. Such unobserved and time-invariant country characteristics for instance include culture or geographic location. Time fixed effects control for unobserved characteristics that influence all countries in the same way but change from period to period. This would typically control for a global financial crisis due to which all countries are equally downgraded, or for years of spectacular global growth leading to a uniform upgrade of all sovereigns. Third, as opposed to Reinhart, Rogoff and Savastano, Banister and Barrot include GDP per capita instead of GDP in their model and expand the number of countries to 120 over 21 years.17 Bannister and Barrot apply the improved model to a sample of non-investment grade Latin American countries to pin down the debt to GDP reduction that would be necessary for each of them to move to improve their sovereign rating.18 For instance, the authors compute how much fiscal consolidation a non-investment grade country should pursue before it can reach investment grade, or how much debt reduction it would take for a highly-speculative grade country to reach speculative. They find that for countries starting with the same credit rating, the fiscal effort they would need to produce to climb to a prime rating can vary considerably.19 This suggests that the market views significant structural political issues surrounding some countries’ ability to borrow. As a result, two countries with the same level of debt to GDP may have very different IIR. This signals that “the market perceives that there are also significant structural issues that affect their level of debt tolerance that would have to be solved to take on higher debt comfortably.”20 Choosing the same IIR target for all countries, the authors compared the country-by-country debt ratios necessary to reach them to build a proxy for debt intolerance.21 The authors conclude their paper by building an index of debt intolerance. Giving the same rating upgrade target to all countries, they work out for each nation the corresponding debt to GDP ratio adjustment that would achieve this upgrade.

5

(3) “The Causes of Sovereign Defaults: Ability to Manage Crises Not Merely Determined by Debt Levels” (Duggar et al. 2010) This Moody’s report identifies factors such as economic resilience, the quality of political institutions, and the structure of the debt that are considered to be equally as important as debt levels when determining a country’s ability to manage crises.22 The report includes findings from a historical study conducted over the 20 sovereign defaults on government bonds that occurred between 1997 through 2010. The report contains four key findings: (1) While defaults are correlated with rising debt burdens, a high debt-to-GDP ratio is neither a necessary nor a sufficient condition for sovereign default. (2) Past defaulters had high foreign exposure, an average of 76 percent of total debt was in foreign currency in the year prior to default, and high debt servicing costs. (3) Sovereigns with moderately low debt levels have defaulted when their economic prospects were poor, their net foreign currency exposures were large, and their political institutions were weak. (4) Conversely, countries with high economic resiliency, debt that is predominantly denominated in domestic currency and strong political institutions have historically been successful in managing relatively large debt burdens and eventually reversing increases in debt-to-GDP ratios caused by macroeconomic shocks and banking crises. The Moody’s report categorizes four main roots in modern sovereign defaults: banking crisis; chronic economic stagnation; high debt burden; and institutional and political factors. In 15 percent of cases, the main cause for the missed payments was a banking crisis, where a banking sector bailout lead public debt to balloon out to unsustainable levels, prompting and a currency crisis.23 For another 15 percent, the country chiefly defaulted on its obligations because of protracted economic stagnation.24 Some of the largest sovereign defaults in modern history debt crises including the 1998 Russian and 2001 Argentine defaults resulted from a combination of many factors including stagnating economic conditions, weak fiscal profiles due to a shrinking base and plummeting revenues, and capital outflows triggering banking and currency crises.25 In 35 percent of cases, the crisis occurred because of unsustainably high debt burdens.26 In such cases, the slow accumulation of debt and progressive erosion of debt affordability, either as a result of ill-advised fiscal expansions or severe external shocks drying up public resources, led to an eventual default at very high debt to GDP ratios. In the remaining 35 percent of cases, institutional and political weaknesses accounted for the country’s default, explained by regime instability, failing debt management/macroeconomic management or even unwillingness to pay. Defaults due to institutional factors historically have lower recovery rates than the three other categories.

(4) “The Institutional Determinants of Debt Intolerance” (Giordano and Tommasino 2011) Economists Giordano and Tommasino expand on the role of political and monetary institutions in debt intolerance in their chapter from the book Sovereign Debt: From Safety to Default. The 6

authors begin with an observation that governments often make a decision to default even if they could afford to pay back debts through other decisions like reducing expenditures or increasing revenues.27 This indicates that institutional factors may ultimately explain debt intolerance.28 They find evidence that there exists a country-specific debt threshold above which defaults occurs, and this threshold increases with the increase of political influence from the middle class and with greater central bank independence.29 Countries which lack adequate political and monetary institutions will suffer debt intolerance. In assessing political and monetary institutions, Giordano and Tommasino work from a political economy perspective in their analysis under the assumption that policy makers are self-interested individuals driven to advance their own political objectives given institutional constraints. With respect to political institutions, the authors believe that high debt situations leave social groups with conflicting economic interests because policies can have widely varying redistributive outcomes.30 For a rational government, the decision to default would only be made if the benefits outweigh the costs on its constituency. Under these assumptions, Giorgano and Tommasino analyze constituents’ interests.31 For example, middle-class households, which experience low interest rates and cash holdings, tend to hold the most public debt and are the most exposed to government bankruptcy. Improving a country’s debt sustainability would depend on the level of political influence this middle class holds.32 Further, the presence of domestic institutions likely protects the interests of the middle class and reduces the chances for individuals in the upper class to unduly wield more power.33 Similarly, monetary institutions play a role in default according to Giordano and Tommasino.34 In most countries, laws establish and protect the independence of a monetary authority from government influence. Checks and balances in the political process reduce the chances that government will override monetary authority. Social groups such as the financial community may also defend central bank independence. Within the central bank, objectives tend to be less partisan given appointments to the bank from more than one branch of government. A central bank is likely to protect the average citizen opposing a pro-wealthy government demands for a loose monetary policy and opposing a tight policy from a pro-poor government. Therefore, in a debt crisis a truly independent central bank will resist political pressure to ensure both rich and poor would share costs of a default. Under these circumstances, a government would have an incentive to honor its debts even without the presence of a powerful middle class. The study draws upon empirical evidence covering income distribution, central bank independence, and features of political institutions for 192 countries.35 They accounted for changes in both political and monetary institutions by splitting dataset into two periods: 1975 to 1990 and 1991 to 2006.36 They found in a cross-section of middle income countries that, on average, income inequality is significantly higher in countries that experienced at least one episode of default.37 And they indeed find evidence of a negative correlation between the size of the middle class and central bank independence on the risk of default. Giordano and Tommasino’s research on the relationship between institutions and debt intolerance supports our decision to include factors of economic freedom in our predictive models.

7

Data

The data used in this study includes data of 66 countries. The map below shows the countries included in the study.

Our first data analysis looks into the relation between debt to GDP ratios and credit ratings within the sample used:

60000 Debt to GDP and Credit Rating Relation

50000 Southern EU China 40000 Upper medium grade GCC

30000

20000 GDP per capita USD) (constantcapita GDP per

10000

0 0 10 20 30 40 50 60 70 80 90 100 Credit Rating *every point represents a five year period The graph shows that there seems to be a break between countries around a threshold of $20,000 of GDP per capita. This is consistent with the Middle Income Trap literature, a theory that analyzes why some countries get held at certain GDP per capita levels. 8

The graph also shows four different groups with particular behaviors. Looking into other categories of countries, China makes a distinct category due to its lower GDP per capita levels with high ratings; a second group can be spotted of Southern European Union countries, with higher wealth levels and lower ratings. A third group is composed of Gulf Cooperation Council (GCC) countries, with high wealth and yet failing to reach high ratings; and the fourth final group shows countries with a GDP per capita levels below $20,000 but that managed to attain relatively high rating levels. The rest of the countries are debt intolerant and debt tolerant countries. The first category are countries with low GDP per capita and low credit rating and the second are countries with high GDP per capita and higher ratings.

100 Debt to Credit Rating

90

80

70 Upper Medium Grade China Southern EU 60 GCC

50

Credit Credit Rating 40

30

20

10

0 0 50 55 100 150 200 250 Debt/GDP

When looking at the relation between debt and ratings, there is also a distinction between tolerant and intolerant countries. Countries in the first category tend to be able to accumulate debt without major impact to their ratings, the opposite happens in intolerant countries. Southern European countries display a similar pattern to intolerant countries, but at higher ratings.

The same applies to countries from the GCC. China shows again a different behavior by being able to increase its ratings with similar debt levels. Finally, the upper medium grade countries that have a GDP per capita of less than $20,000 share with debt intolerant countries the need to keep low levels of debt to maintain better credit ratings. In our sample they remain at debt levels below 55 percent of GDP.

9

100 Credit Rating to Index of Economic Freedom 90

80 Debt Tolerance 70

60

50

Credit Credit Rating 40

30 Southern EU 20 GCC China 10 Upper Medium Grade

0 20 30 40 50 60 70 80 90 100 Index of Econmic Freedom *every point represents a five year period When looking at the relation between the Index of Economic Freedom (IEF) and ratings, a clear positive relation between institutionality levels and ratings for debt intolerant countries can be seen: while debt tolerant countries show minimum levels of 60 of the IEF with the majority concentrated around a level of 70. China shows again a distinctive behavior, showing higher ratings compared to other countries in a similar situation. Finally, Southern European Union countries and countries from the GCC generally show lower levels of economic freedom compared to debt tolerant countries. The same situation applies to countries with lower than the GDP per capita threshold and high ratings.

To build this analysis, we gathered data from different sources, including the IMF, World Bank, Heritage Foundation, Institutional Investor and Moody’s. The table below describes variables included in the study:

Variable Definition Unit of Measurement

Inflation Dummy 1 if inflation > 10 = 1 or else 0 0-1

Debt to GDP Ratio Total public debt/GDP Percentage out of 100

GDP per Capita GDP per capita in constant 2010 U.S. dollars. U.S. Dollars

Default Average Lagged The default average in previous time period (t-1) 0-1

IIR Lagged IIR in previous time period (t-1) 0-100

Index of Economic Freedom Score in the Index of Economic Freedom 0-100

Time Periods Time periods of 4-5 years 1-6 *For further information about the variables please refer to the Appendix.

10

Explanation of Our Model

This section of the report explains the econometric models we used to estimate debt intolerance. Building on Bannister and Barrot’s revised Reinhart, Rogoff and Savastano model from 2011 as a foundation, we estimated two equations for debt intolerance using a standard fixed effects approach. In the first equation (“Model 1”), we regressed the IIR on a number of indicators of debt intolerance, including the history of a country’s default and debt level, to put forth the marginal effect of each indicator on the ratings and the debt intolerance. In the second equation (“Model 2”), we replaced the variable for default history with a variable that measures institutional strength to explore the effect of this variable on ratings as suggested by some of the literature. Using the models, we estimated debt intolerance equations showing the country-specific functional relationship between the IIR and a set of variables including the debt levels.

Both models presented improve on the previous studies by using panel data for more recent periods, incorporating both historical and contemporaneous variables, using different functional forms of variables. Model 2 replaces the history of default with an institutional strength variable based on the Index of Economic Freedom. Since the institutional strength is an indicator of the capacity of the government to conduct sound policies that foster and sustain , we used the Index of Economic Freedom as a proxy of concurrent strength of the institutions, instead of default history as a proxy of chronic institutional weaknesses.

Model 1 Model 1 uses panel data for 66 countries spanning 25 years between 1990 and 2015 and the second uses panel data for the same countries spanning 15 years between 2000 and 2015 (Appendix Table 3). The country and time period selections were based on the availability of data. As Reinhart, Rogoff and Savastano’s database provides a time series of the external default for 70 countries (with a population over 1 million), we restricted our models to those countries and excluded some of the countries due to lack of reliable data on the other variables. The purpose of this methodology is to create a balanced panel. Moreover, as the IIR covers the countries in the sample for the specified time period and highly correlates with Moody’s ratings, the IIR is used as dependent variables in the models as a proxy for sovereign default risk.

2 Credit = ß0 + ß1 CRit-1 + ß 2 Debt to GDPit + ß3 Debt to GDP it + ß 4 GDP per capitait + Ratingit ß5 Dummy Inflationit + ß6 Default Averageit-1 + ß7 Time Trendt + Country Fixed Effecti + uit

The debt intolerance equation in Model 1 is estimated by regressing the IIR on (1) previous period’s IIR, (2) ratio of debt-to-GDP, (3) ratio of debt-to-GDP squared, (4) level of GDP per capita, (5) a dummy for periods of inflation over 10 percent, (6) lagged default dummy showing proportion of years with or default within each period, (7) country-specific fixed effect and (8) time fixed effect. The previous period’s IIR is used to capture the dependency of the current rating on the preceding rating and the gradual changes in the level of the IIR except the periods of default. The ratio of general -to-GDP as a proxy of the fiscal strength

11

is a key factor in the default risk. The ratio of debt-to-GDP squared is also used to control for the non-linear effects of debt on the IIR. While the lagged default dummy is used to capture the historic institutional weaknesses of the country, the inflation dummy is used to capture the concurrent macroeconomic instability of the country. The level of GDP per capita as a proxy for economic strength is used to control for the level of economic development of the countries. Moreover, by using country fixed effect and time fixed effect panel data approach, the model controls for the unobserved characteristics of the countries fixed over time and time trends in IIRs of the selected sample of countries. By controlling for the aforementioned variables, the model measures the marginal effect of an additional unit of debt on the IIR and hence on debt intolerance.

Model 2

2 Credit = ß0 + ß1 CRit-1 + ß 2 Debt to GDPit + ß3 Debt to GDP it + ß 4 GDP per capitait + Ratingit ß5 Dummy Inflationit + ß6 Institutional Strengthit + ß7 Time Trendt + Country Fixed Effecti + vit

The second debt intolerance equation presented in Model 2 is estimated by regressing the IIR on the same variables as the first model except the lagged default dummy. The lagged default dummy is replaced by the Index of Economic Freedom. The Index of Economic Freedom quantifies economic freedom based on 12 quantitative and qualitative factors, grouped into four broad categories of economic freedom which are (1) the rule of law, (2) government size, (3) regulatory efficiency, and (4) open markets.38 These factors function as a proxy for a country’s institutional strength.39 While there are several others indices that measure institutional strength such as World Governance Index and Polity, we applied the Index of Economic Freedom based on the ease of including its data given our time constraints. However, the models could also benefit in the future by including these others indices. Finally, as economic freedom and a functioning market economy have become increasingly discussed in the context of steady economic development, the sensitivity of the ratings to the Index of Economic Freedom is studied in this model to clarify the contemporary debt intolerance particularly in the emerging markets.

12

These are the coefficients that explain the model:

Table 1. Estimation of the Debt Intolerance Equation Model Specification Model I Model 2 Inflation Dummy -3.647*** -2.902** [-0.9660] [-1.2090] Debt-to-GDP Ratio -0.224*** -0.245*** [-0.0297] -0.0430] Debt-to-GDP Ratio Squared 0.000618*** 0.000625** [-0.0001] [-0.0002] GDP per Capita 0.000680*** 0.000719** [-0.0002] [-0.0004] Default Average Lagged -3.770** - [-1.5210] IIR Lagged 0.508*** 0.467*** [-0.0585] [-0.0810] Index of Economic Freedom - 0.204*** [-0.0700] Constant 25.55*** 18.42** [-6.0020] [-8.9020] Observations 330 264 R-squared 0.973 0.973 Robust standard errors in brackets Statistical Significance levels: *** 99 %, ** 95 %, * 90%

13

Data Analysis: Latin America Going again into the data analyzed, this paper tries to explain how countries from the region can improve their ratings. The table below compares IIR and Moody’s ratings. Targets will be placed according to the rating description categories as ratings may differ for specific countries between IIR and Moody’s but not on their assigned categories1. For example, a country like Germany with an IIR of 94 would be placed on the rating category of High Grade and will have as its target Prime with a value of 98.

Moody's Rating Assigned Value IIR Rating Category Aaa 21 96-100 Prime Aa1 20 92 -96 Aa2 19 88 -92 High grade Aa3 18 84 -88 A1 17 80 -84 A2 16 76 -80 Upper medium grade A3 15 72 -76 Baa1 14 68 -72 Baa2 13 64 -68 Lower medium grade Baa3 12 58 -64 Ba1 11 52 -58 Non-investment grade - 10 44 -52 Ba2 speculative Ba3 9 40 -44 B1 8 36 -40 B2 7 32 -36 Highly speculative B3 6 28 -32 Caa1 5 24 -28 Substantial risks Caa2 4 20 -24 Extremely speculative Caa3 3 16 -20 Default imminent with little Ca 2 12 -16 prospect for recovery C 1 0 -12 In default

The table below places countries according to their current rating category and sets a debt target for their rating improvement.

1 For further information about the comparison between IIR and Moody’s ratings please refer to the Appendix. 14

The first group is composed of developed countries, which manage to sustain high debt levels without seeing their ratings affected. Among the Latin American countries there seems to be three levels of debt intolerance with corresponding IRR levels, clearly illustrated in the graph below.

100 DEU USA 90 FRA 80 MEX CHL 70 ESP BRA COL 60 PER PAN CRI URY 50 PRY Credit Credit Rating 40 DOM GTM BOL SLV 30 VEN HND ECU NIC 20 ARG

10

0 0 10000 20000 30000 40000 50000 60000 GDP per capita

The first one are countries with comparatively higher ratings, with a ceiling of 76, and it includes Chile and Mexico. The second group is composed of investment grade countries that only achieve a maximum IIR of 64, which is close to the threshold for the investment grade. Brazil, Peru, Uruguay, Panama, Costa Rica and are in this group. The rest of the countries cannot achieve investment grade even at the expense of having no debt, their GDP per capita levels or IEF levels is so low that it greatly affects their debt tolerance. If we do the same exercise taking into consideration the IEF as a reference of an institutional index, the new target are:

15

The model now suggest that countries that already have an investment grade category do not seem to be able to improve their ratings by improving their institutions. However, countries with low ratings could get a great rating improvement if they manage to rank higher in the institutional index. In this case two clear groups can be identified: countries with relatively weak institutions (IEF of less than 60) and countries with relatively stronger institutions. Countries with weak institutions and ratings have the most to profit from improving their ratings. These countries can improve their ratings without seeing a change in their GDP per capita, with the exception of Nicaragua that misses the threshold by one point.

Finally, we reduced debt levels to 0 and improving the Index of Economic Freedom (IEF) to 100 and compared the results. The observed IIR for the year 2015 is on the horizontal axis; the vertical axis reflects the maximum possible rating that can be achieved without a change in other variables (other than debt and economic freedom levels).

100 USA, 100 MEX, 89 90 BRA, 87 CHL, 86 80

70 (Debt (Debt & Eco Free)

60

50

Max Credit Rating Max Credit Rating 40

30 63 71 77 93.6 20 30 40 50 60 70 80 90 100 2015 Credit Rating

Our case studies, Mexico, Brazil and Chile, won’t achieve a 100 score, even without debt. This could show some debt intolerance, especially when compared with the US. It could also show the limitations of our model and the complexity of measuring debt intolerance. 16

We can identify from the analysis three groups among the Latin American countries described on this section. First, there seems to be a ceiling of IIR around 76 for all Latin American countries, only achievable at low debt levels and having GDP per capita levels above ten thousand dollars and IEF of sixty or above. The second group are countries that have usually higher debt levels and seven thousand dollars or, as in the case of Peru, lower GDP per capita and lower debt levels. The final group includes countries with lower IEF levels and GDP per capita on average, with the exception of and . These two show the lowest IEF levels on the series and, therefore, lower IIR ratings than what would be expected when looking at their GDP per capita and debt levels.

Case Study: Mexico

Moody’s Sovereign Credit Rating: A3/negative In 2016, Mexico risked a cut in its credit rating as the three main rating agencies, including Moody’s, revised the country’s outlook from stable to negative.40 Key reasons for this revision included lower economic growth, low oil prices, rising debt and interest burden, and uncertainty about future trade relations.41 Leading up to this negative outlook Mexico’s gross debt to GDP has been increasing from 39 percent in 2005 to 52 percent in 2015.42 Analysts maintain this negative outlook today as Mexico continues to address these challenges. The model we recreated considers some of these concerns identified by analysts. A small percent of the rating can be explained from factors evaluated in the Index of Economic Freedom. Accordingly, this case study begins by identifying and analyzing some of Mexico’s weaknesses under this index’s two components of rule of law: (1) judicial effectiveness, (2) government integrity.

Data Source: “Index Economic Freedom.” The Heritage Foundation, April 8, 2017 Judicial Effectiveness Judicial effectiveness requires protection of rights of all citizens, and efficient and fair judicial systems to ensure laws are fully respected.43 According to the Index of Economic Freedom,

17

Mexico’s judicial effectiveness is extremely low with a score of 38.7 out of 100.44 Mexico’s score falls in the same range as the average for Latin America and well below developed economies such as the US. A strong judicial system could provide the basis for stronger economic growth. For example, studies in Mexico have shown that large companies are more inclined to invest in states with better court systems.45 This is because better courts reduce the risks firms face.46 Government Integrity In terms of government integrity, Mexico ranks moderately below the average for Latin America with a score of 30, but has been progressively worsening since 2005. Government integrity primarily focuses on elements of corruption. Corruption costs Mexico annually between 2 percent and 10 percent of its GDP, while reducing investment by 5 percent and eliminating 480,000 jobs in small to medium size businesses.47

Model 2 Implication for an Improvement in Economic Freedom The model suggests that significant improvements to Mexico’s scores in judicial effectiveness and government integrity along with other improvements in labor freedom could encourage a higher rating. Many of these factors are indirectly related to economic growth, one of the key factors cited by analysts for the 2016 negative outlook.

Implications about the Ratings Under Our Models Despite moderate levels of public debt, Mexico has a lower credit rating compared to the advanced countries with higher levels of public debt. According to our models, Mexico can upgrade either by a substantial decrease in its public debt or by a remarkable improvement in its scores under the Index of Economic Freedom. In fact, the debt intolerance of the Mexico results both from the level of economic growth and institutional weaknesses. According to our sensitivity analysis, it appears important for Mexico to sustain economic growth while ensuring greater fiscal discipline to maintain the current rating as its rating is close to the estimated lower rating thresholds. The models suggest a small decrease in GDP per capita and further increase in debt in 2016 would have put downward pressure on the rating. In this respect, the models’ implications are consistent with the Moody’s rationale for the outlook change from stable to negative in March 2016, which was primarily due to slower than expected growth and risk of increasing debt burden as well as inefficiencies in state-owned oil producer PEMEX.48 Thus, Mexico has relatively low debt tolerance and it has limited fiscal space to tolerate low growth prospects and institutional weaknesses.

18

A Break from the Past? Mexico’s past two episodes of severe financial stress in 1982 and 1994 followed periods of heavy borrowing. The first instance occurred in 1982 following an oil shock that sent global oil prices tumbling. Mexico, which had recently discovered enormous petroleum deposits, had been borrowing excessively against future revenues.49 As the US Federal Reserve increased interest rates, a period of capital flight followed, and by August 1982 the government could not pay its debt which had grown substantially.50 This event triggered a wave of debt crises throughout the region. By 1989, a series of deals from the US negotiated under the Brady plan targeted debt- forgiveness leading with Mexico, which helped alleviate some of the stress.51 Just five years later in 1994, Mexico experienced a second instance of financial stress culminating in the “Tequila Crisis.” The lead up to the crises involved a series of economic reforms including trade liberalization, especially NAFTA, and liberalization of capital flows.52 NAFTA opened Mexico to foreign direct investment. This time political instability, pegging the peso to the dollar and issuing short term peso treasury bills promising repayment in US dollars known as “Tesobonos” played a role in default.53 By 2003, Mexico had become the first country to pay off $35 billion in its Brady bonds.54

Data Sources: British Petroleum, International Monetary Fund and Bureau of Labor Statistics, April 14, 2017. As the graph above illustrates, Mexico may not have broken from the past. Despite an auto manufacturing industry which has soared since the implementation of NAFTA and 9 other free trade agreements, Mexico, nonetheless, remains vulnerable to oil shocks.55 About 30 percent of 19

the government's revenue is from oil.56 Mexico has increased its debt to GDP ratio from 33 percent in 2005 to 43 percent in 2015. At the same time, both domestic oil production is down 33 percent and global oil prices are down more than 50 percent. This has led to lower oil revenue while Mexico’s economy experienced slower economic growth. The government has not significantly diversified revenue sources in the past 25 years. Tax revenue still only makes up 15 percent of total GDP compared to over 25 in the US and over 30 percent in other developed economies.57 Capital flight remains a real risk as foreigners have increased their holdings of local currency government bonds to 35 percent, more than any other emerging market country.58 According to Reinhart and Rogoff sudden and substantial capital inflow can place countries at a high risk of debt crises.59 These conditions in their totality leave Mexico vulnerable to shocks from abroad and capital outflow,60 which leaves the country in familiar territory.

Case Study: Chile

Moody’s Sovereign Credit Rating: Aa3/stable As of April 2017, Chile has the highest ratings among the Latin American countries and is the only Latin American country rated in the “A-zone” by Moody’s and other rating agencies. Moody’s July 2016 evaluation predicts the country will “retain a strong fiscal position despite increases to the main debt metrics and a slowing of the economy.”61 The report contains a positive outlook despite expectations of growing debt, explaining that “although Chile's debt burden will rise this year and next, it will do so slowly and remain significantly lower than most rated peers.”62

Chile’s rating is influenced by two factors: (1) strong institutional indicators; and (2) small debt to GDP ratio. The July 2016 said institutional framework was “a key ratings support for many years,” and that it “remains strong despite political opposition to and investor concerns about President Bachelet’s policies.”63 These findings appear to be supported by our model, where Chile can explain an entire notch of its rating due to its institutional framework, making it one of the most successful countries in this area.

Chilean Economic Transformation The Chilean economy has undergone significant transformations following reforms initiated in the mid-1970s and it also experienced a substantial economic boom in early 1990s.64 The long-term economic success in Chile can be explained by the four pillars of its macroeconomic framework: a structural fiscal rule; price stability through inflation targeting and a flexible exchange-rate regime; sound financial system; and open trade and financial integration to the global markets. Claudio Sapelli, a Chilean economics researcher, analyzed the democratic transition and found that the Chilean institutional structure for the transition was crucial for its economic success. He explains, “The Chilean process is remarkable in its concern to appropriately design both big and small institutions so to provide an adequate framework for decision making by the private sector.”65

20

Data Source: “Index Economic Freedom.” The Heritage Foundation, April 8, 2017 The Index of Economic Freedom ranks Chile in 2016 as the best country in Latin America and one of the best in the world. Chile performs better than Latin America and the US in fiscal health, and this includes the fiscal rule that the government must fulfill. The tax burden is better than in the rest of Latin America. Other than government spending, Chile performs better than the Latin American average.

The second factor that can best explain Chile’s rating is its debt to GDP level. The country has managed to grow at an above-average rate for the last 30 years, when compared to the rest Latin America. From 1990 to 1998 Chile managed to grow its GDP at an average rate of over 7 percent. From 1999 to 2015 the GDP of the country grew at an average rate of 3.6 percent, still higher than the Latin American average of 2.76 percent. The growth was also accompanied by relatively low fiscal deficits and fiscal surplus in economic boom periods, creating a buffer for harder times.66

Data Source: International Monetary Fund, World Economic Outlook Database, October 2016.

21

Implications about the Ratings Under Our Models Chile has lower credit rating compared to advanced countries with higher levels of public debt despite its remarkable macroeconomic and fiscal performance as well as better institutional strength indicators. According to the models, Chile can not improve its ratings only by decreasing its public debt and it can upgrade only by a quite large improvement in the Index of Economic Freedom. The models suggest Chile needs to improve its GDP per capita substantially, and, therefore, needs to expedite its economic growth. Considering that Chilean total factor productivity growth has been negative in the last two decades, Chile would likely need to implement structural reforms to boost its total factor productivity in the long run.67 However, our sensitivity analysis suggests Chile has a relatively low debt tolerance compared to more advanced economies and it has limited fiscal space to finance required structural reforms.

Macroeconomic and Fiscal Outlook Despite rapid recovery and stunning macroeconomic performance after the global financial crisis, the GDP growth rate of Chile has decelerated since 2014. In fact, the annual growth averaged 2.1 percent since 2014. Despite this, the Chilean GDP growth is expected to remain above the average of economic growth of countries in the region.68

The slowdown in Chilean economy has been attributed to external as well as domestic challenges.69 The weakening external demand, mainly due to slower growth or recession on its main trading partners Brazil and China and continuing copper price declines has been flagging the Chilean exports. Exports in Chile highly depend on copper, which accounts for more than 45 percent of exports.70 Due to the adverse effects of uncertainties of the last government’s reform agenda on consumers as well as investors, domestic demand has been undermined. In addition to 22

the recent external and domestic challenges, structural weaknesses - including low human capital, infrastructure bottlenecks, an aging population and a high dependency on copper - limit the potential for long-term economic growth. According to the last IMF article IV report, the main domestic risks are “a delayed recovery in business confidence and investment related to larger the expected uncertainties surrounding a new labor bill.”71 Uncertainty could also come from the pension system which was the subject of flagship reforms in the 1980s that allowed the country to control its social expending. The government now faces growing social demands to change the private pension system. A new system could increase contribution rates that, according to the IMF, could dampen economic growth over the medium term.72

Chile’s countercyclical policies applied in the last two decades have been widely acknowledged as a model for a commodity export country. However, problems may arise in the upcoming future. Chile managed to create a sovereign fund and increased expending when the economy was not doing as well as before. A major issue will focus on the outcome if copper prices do not reach previous levels. In that case, the country may have to reduce spending while social demands are growing. To prepare for this possibility, the current government of Bachelet adopted policies for free education and tax reform. Nonetheless, the debate seems to be increasingly focusing on the pension system, one of the crucial pillars of the Chilean model and the fiscal discipline of the country.

Case Study: Brazil

Moody’s Sovereign Credit Rating: Ba2/neutral Brazil is the largest economy in the region and its GDP represents about half of the Latin American economy. In March 2017, Moody’s raised the outlook of Brazil’s sovereign Ba2 rating to stable, from negative.73 The key drivers behind this decision include a decline in downside risks and stabilization of macroeconomic conditions, improvement in the country’s policy framework, and a reduction in the risk of contingent liabilities from government-related entities. From these three factors, the amelioration in the policy framework seems to carry the biggest weight in changing Moody’s expectations according to the agency’s internal analysis.74 As for the country’s performance in the Index of Economic Freedom, the country has fallen after previous gains in the list and continues to lag behind the regional and world average.75 Factors such as the government size, regulatory efficiency and market liberalization are the areas where Brazil ranks comparatively worse.

23

Data Source: “Index Economic Freedom.” The Heritage Foundation, April 19, 2017 Fiscal outlook After a decade of generous growth, Brazil’s economy slipped into recession in 2014, and has since recorded stagnant or falling GDP. The slowdown of the Chinese economy, a long-overvalued real, structurally high production costs, renewed inflation and ongoing political turmoil shook the country out of its bonanza (2003 – 2014), sending growth from a 10 percent high in 2009 to about -3 percent in 2015.

This downturn – the latest bust in a series of business cycles that shaped the Brazilian economy since the 1960s – deteriorated the country’s debt profile. Specifically, imprudent fiscal policies, including costly price control of utilities designed to stave-off inflation, and high domestic interest rates have led to successive fiscal deficits. In its 2016 Article IV, the IMF points out that gross debt of the nonfinancial public sector (NFPS) increased by 10.5 percentage points of GDP in 2015, reaching 73.7 percent of GDP. Net debt of the public sector rose by only 3 percentage points of GDP.76 Foreign holdings of government securities has remained broadly constant in percent of GDP (16.6 percent). However, net interest on NFPS debt picked up significantly in 2015 due to more elevated borrowing costs and because of the large losses on the FX swaps. Primary surpluses above 3 percent of GDP, excluding interest revenue, would be needed to keep gross debt from growing as a ratio to GDP beyond the IMF Article IV debt sustainability analysis projection horizon.

Nevertheless, due to the Brazilian government’s unique debt management scheme, Brazil’s debt burden appears sustainable. The federal government domestic tradable securities accounted for 90 percent of total NFPS gross debt in 2015, of which approximately two-thirds were held by the public. Active management in recent year has improved the average maturity of these instruments, which now is 4.6 years from 3 years in 2008. However, nearly 22 percent of securities matured within the last year. In response, the government has as an objective to raise the average maturity to 5.5 years and to bring the short-term debt down to 20 percent of the total debt over the medium term. The challenge here is how to deal with the historic high level of interest rates Brazil has experienced.77 The Article IV report further notes fixed-rate and inflation-linked domestic bonds have gradually replaced foreign-currency linked instruments and floating-rate bonds over the last 24

half a decade. Zero-coupon bonds with original maturities over one year constitute slightly more than half of federal government domestic tradable securities held by the public, or 21.7 percent of GDP. Foreign holding of domestic securities was stable at 18 percent of total at end-2015.

Implications about the Ratings Under Our Models Although Brazil has lower rating compared to the advanced countries with similar public debt levels, it is still more debt tolerant compared to some other Latin American countries (e.g. Colombia, Uruguay) as it manages to have higher debt with similar ratings. According to our models, the debt intolerance of Brazil results both from the level of economic growth and institutional weaknesses. The sensitivity analysis suggests that although Brazil can improve its rating by a moderate decrease in its public debt, it needs a significant improvement in the Index of Economic Freedom or substantial economic growth to catch other upper grade countries in the region.

Throughout the period, 2000 – 2015, Brazil’s debt to GDP ratio has a constant impact of about - 15 on the IIR. This is considerably more than Chile (about -3) but in line with Mexico and the US (about -18 throughout the period). The impact of economic freedom on Brazil’s IIR is about constant, and in line with our other case studies.

Interestingly, it appears that between 1990 and 2005, the market overestimate Brazil’s sovereign risk, awarding it an average 37.14, 40.33 and 50.79 IIR for 1990-1995, 1995-2000 and 2000-2005, against predictions by our traditional model of 40.97, 45.88 and 52.99. This trend then reverses in the subsequent decade, with a positive discrepancy (underestimation of sovereign risk) of 7.78 in 2005-2010 and 3.81 over 2010-2015. These discrepancies confirm Reinhart and Rogoff’s 2003 finding that, when assessing sovereign risk, markets overreact (both positively and negatively) to the macroeconomic environment. This suggests low investor confidence following the Real Plan 25

and, coincidentally, during the Asian financial crisis, and overconfidence in the boom years and the post-crisis growth peak.

Break from the past? If one wants to understand Brazil’s most recent recession, we can look at past policies to explain the country’s recent crisis. However, the current crisis has also some unique attributes

First and foremost, surging inflation since 2015 is reminiscent of Brazil’s earlier struggles with price levels. The inflation rate has exceeded the 4.5 percent target since 2009 and shot to 10.7 percent in 2015, falling to 8.7 percent last year. Prices have been pushed up by supply shocks (most notably droughts pressuring food and energy prices), utility price adjustments (the government lifting energy prices to make up for years of underpricing) and indexation.78 Inflation has also been fueled by the progressive devaluation of the real over the past years. The central bank has renewed its commitment to inflation, which is expected to be below target (4.3 percent) by 2019. This trend pales in comparison to the 1,000 percent inflation witnessed in the 1990s, but has nonetheless antagonized those electors who remember Brazil’s lost decade.79

Although former President Dilma Rousseff’s austerity measures have received considerable attention for their potential impact on growth, the recession is very much a reflection of Brazil’s past policy choices. Most notably, the industrial sector has been decimated by the Real Plan (1994) and the subsequent overvaluation of the currency during the early 2000s. Brazilian growth was mostly fueled by China’s economic expansion, and the latter’s slowdown, in conjunction with high domestic interest rates, has left the Brazilian economy struggling to recover. Dilma Rousseff’s fiscal consolidation only had a limited impact on growth upon announcement by the appointed finance minister in December 2014, but got diluted among the political chaos.80

The crisis’s unique character precisely comes from the current political turmoil, that saw Dilma Rousseff impeached under the suspicion that she downplayed the government’s fiscal deficit. The “Lava Jato” investigations – a corruption probe revolving around Petrobras – have not abated since. In December 2016, 77 executives from Brazil’s largest construction company stepped down, indirectly implicating the president and several ministers and legislators. Until the end-2018 elections and the end of investigations, building and sustaining a majority in congress will prove hard.81 Despite such adverse circumstances, the current President Temer has pursued his reform agenda, imposing a constitutional amendment to freeze public expenditures.

Case Study: US

Moody’s Sovereign Credit Rating: Aaa/stable In 2011, following the Debt Ceiling Crisis, Standard and Poor’s became the first and only major US rating company in history to downgrade the US rating. Meanwhile, many other ratings companies put the US on a negative watch. Analysts cited several factors for the downgrade, but broadly speaking it reflected views of a weakening in the effectiveness, stability and predictability of US policy making and political institutions during a time of economic and financial struggle.82 Raising the debt limit has been a recurring issue in Congress for decades,83 but this time, Analysts 26

viewed this particular instance with skepticism. Following the rise of the conservative Tea Party movement and the 2010 midterm elections in which the Republican party took the house of Representatives, the debt ceiling became a political bargaining chip in the debate over fiscal policy. Though S&P was the only major US rating company to downgrade the credit rating, other analysts had placed the US on a negative watch. Some analysts have not upgraded the US despite the removal of the negative watch by others. Undoubtedly, the US has enjoyed a long history of debt tolerance far beyond emerging market countries in large part due to: (1) a large and diverse economy, (2) reliable sources of government revenue, (3) how the debt is used, and (4) strong economic and political institutions.

Large and Diverse Economy The US has been the world’s leading economic power for over 140 years, and on average has accounted for one third of the world’s GDP since 1960.84 Technological innovation has been a key driving factor for increased productivity in this period and the US continues to dominate in technological advancement across many sectors. The US economy is diverse, reducing the impact from external events. For example, in 2009 despite the size of Subprime Crises industries outside of the banking sector such as technology were able to pick up the pieces and move the economy forward. Reliable Sources of Government Revenue Throughout most of its history, the US government has collected reliable sources of revenue made up almost entirely of tax. For example, in 2014, the US government collected over US $3 trillion in , enough to cover nearly all its expenses.85 Since 1950, almost half of the government revenue came from income tax and one third from payroll taxes.86 This helps the government manage its debt levels without sudden needs for increased borrowing as may be the case when an emerging country relies on less reliable sources of revenue such as commodities.

How Debt is Used The purpose behind debt build up is important. When the US government has run large budget deficits, funded by debt issuance, it has mainly been to finance war (see figure below). Reinhart and Rogoff argue that war debts are less problematic for a country’s economic growth than large debts accumulated during peacetime.87 This is due to military spending that also benefits the civilian economy. Furthermore, debt build during wartime is expected to be temporary.88 In comparison, large borrowing in other times reflects deeper economic and political problems that may persist without a foreseeable end. 89

27

Strong Economic and Political Institutions Under the Index of Economic Freedom, the US ranks in the top 20 countries with strengths in rule of law, regulatory efficiency and open markets. However, the index notes decline in economic freedom when considering rising large budget deficits and a high level of public debt.90 The US demonstrates strong inclusive economic institutions with secure property rights, law and order, good business climate, access to education and opportunity for social mobility.91 With the strong protection of property and contract rights and open markets this creates an environment for investment and innovation. The US has a long history of inclusive political institutions with pluralism, checks and balances and rule of law.92 In this backdrop of strong political institutions, raising the debt limit has become an issue over the last several decades.93 Nonetheless, members of Congress and the President know all too well the catastrophic consequences of not paying the bills. Although in recent years political will has been delayed, it always seems to resurface in the 11th hour. The US government’s policies on the debt ceiling, therefore, remain reasonably predictable.

28

Conclusion

This report makes the following findings:

 The model we recreated and modified to include a measure institutional strength still does not fully explain the debt intolerance phenomenon in Latin American and other countries. The results suggest there are other variables that play a role and that case specific validations are still necessary to understand a full picture within a country.

 Consistent with the debt intolerance literature and the methodologies used by the rating agencies, the analyses showed that the macroeconomic and fiscal stability, default history, levels of economic development as well as institutional strengths of the countries are the key factors in explaining the debt intolerance and the sovereign credit ratings.

 The sensitivity analysis suggests the debt intolerance of Latin American countries compared to advanced countries results primarily from a low level of GDP per capita as well as institutional weaknesses.

 Despite its remarkable institutional development, Chile still has debt intolerance compared to advanced economies and this is a result of its low economic development as well as its moderate long-term growth prospects.

 Latin America’s score in judicial effectiveness and government integrity is well below developed economies such as the US. Research indicates a strong judicial system and low levels of corruption could provide the basis for stronger economic growth.

 Debt intolerance is a major systematic phenomenon among Latin American countries. This paper shows that institutions play a role in debt tolerance, but we also illustrate that to improve this issue some Latin American countries couldn’t just advance in one variable to drastically improve their tolerance compared with developed countries.

 Our model suggest that, even with substantial improvement in institutions or GDP growth, Latin American countries would remain with comparatively lower ratings than developed countries. We also found that this phenomenon is linked to a certain threshold of about $20,000 of GDP per capita, though further research could go deeper into this.

 Using macro and institutional variables, we can predict future sovereign ratings with reasonable precision and observe the fiscal adjustment necessary for a country to upgrade be upgraded. However, our model is not so effective in discerning the tolerance of countries with low debt to GDP levels.

29

 The market seems to give both Brazil and Mexico higher ratings from what would be expected from our model findings. The gap between observed and predicted credit ratings goes in that direction: in Brazil the gap grew from 2.67 (less than a notch) in 1995 – 2000, to 16.01 (four notches) between 2005 and 2010. This was then followed by a downgrade to non-investment grade in the 2010 – 2015 period. In Mexico, the gap reached 12.38 in 2010 – 2015. Further research could identify whether such overconfidence gaps play any role in predicting rating downgrades.

 Institutions matter for ratings, but improvements on them matter the most below the 60 threshold that is related to lower levels of economic freedom. The major improvements on ratings are observed when countries move from the lower levels towards the 60, beyond that there does not seem to be a clear effect.

30

Appendix

Table 1: Ratings Description

Table 2. Information about the Variables

Variable Definition Unit of Explanation Sources Measurement

Inflation 1 if inflation > 10 = 0-1 If there is a year with inflation is greater than or equal International Financial Statistics of the IMF Dummy 1 or else 0 to 10 % in the specified period, the inflation dummy http://data.imf.org/?sk=5DABAFF2-C5AD-4D27-A175- gets value of 1; otherwise it gets value of 0. 1253419C02D1&sid=1450715373824&ss=1409151240976

Debt to GDP Total public Percentage out Debt to GDP ratio is the arithmetic mean of the annual The Historical Public Debt Database of the IMF Ratio debt/GDP of 100 debt to GDP in each period. http://data.imf.org/?sk=806ED027-520D-497F-9052- 63EC199F5E63&sId=1390030341854

GDP per GDP per capita in U.S. Dollars GDP per capita is the arithmetic mean of the annual World Development Indicators of the World Bank Capita constant 2010 U.S. GDP per capita in each period. http://data.worldbank.org/data-catalog/world-development- dollars. indicators

31

Default The default average 0-1 The sovereign default on external debt is the Global Crises Data by Country of Carmen Reinhart (with her Average in previous time failure to meet a principal or interest payment on coauthors Ken Rogoff, Christoph Trebesch, and Vincent Lagged period (t-1) the due date (or within specified grace period) and Reinhart) includes instances where rescheduled debt is http://www.hbs.edu/faculty/initiatives/behavioral-finance-and- ultimately extinguished in terms less favorable financial-stability/Pages/global.aspx than the original obligation. The default average is estimated by dividing number of defaults in each period by number of years in each period.

IIR Lagged IIR in previous time 0-100 IIR reflects the Institutional Investor's Country Credit Institutional Investor's Country Credit Ratings period (t-1) Ratings which grade each country on a scale of 0 to http://www.institutionalinvestor.com/Global- 100, with 100 representing the least likelihood of Ranking.html#/.WPk7-2nyvup default. IIR is the arithmetic mean of the end of year IIR in each period.

Index of Score in the Index 0-100 Index of Economic Freedom reflects the annual index Index of Economic Freedom Economic of Economic created by The Heritage Foundation and The Wall http://www.heritage.org/index/explore Freedom Freedom Street Journal measure the degree of economic freedom in the world's nations, with 100 representing the highest rank. Index of Economic Freedom value is the arithmetic mean of annual index in each period.

Time Periods Time periods of 4-5 1-6 1: 1990-1994 years 2: 1995-1999 3: 2000-2003 4: 2004-2007 5: 2008-2011 6: 2012-2015

Table 3. Countries in the Sample

1 Algeria 34

2 Angola 35 Mauritius

3 Argentina 36 Mexico

4 Australia 37 Morocco

5 Austria 38 Myanmar

6 Belgium 39 Netherlands

7 Bolivia 40 New Zealand

8 Brazil 41 Nicaragua

9 Canada 42 Nigeria

10 Chile 43 Norway

11 China 44 Panama

12 Colombia 45 Paraguay

32

13 Costa Rica 46 Peru

14 Côte d'Ivoire 47 Philippines

15 Denmark 48 Poland

16 Dominican Republic 49 Portugal

17 Ecuador 50 Romania

18 Egypt 51 Saudi Arabia

19 El Salvador 52

20 Finland 53 South Africa

21 France 54 Spain

22 Germany 55 Sri Lanka

23 Ghana 56 Sweden

24 Greece 57 Switzerland

25 Guatemala 58

26 Honduras 59 Tunisia

27 Hungary 60

28 61 United Kingdom

29 Indonesia 62 United States of America

30 Italy 63 Uruguay

31 Japan 64 Venezuela

32 Kenya 65 Zambia

33 Kuwait 66 Zimbabwe

Estimation Results The country and time fixed effect model estimations with panel data allowed us to control for the time invariant characteristics of the countries that cannot be observed or measured (e.g. cultural factors, national characteristics) as well as global time trends that cannot be measured but affect countries in a similar manner. Table 4. presents the estimation results of the country and time fixed effect panel data regression models. The coefficients in the both models are significant and of the expected sign: ● The positive and large coefficient of lagged IIR supports the persistence of the level of the IIR over time. 33

● The negative coefficient of the default history dummy as a proxy of historical institutional weaknesses shows the adverse effects of the default on the creditworthiness of the country. ● The coefficient of debt to GDP ratio shows the negative association and significant sensitiveness of the IIR to the general government debt level. ● The positive coefficient of debt to GDP squared shows the nonlinear relationship between IIR and the general government debt level ● The positive coefficient of the GDP per capita shows the positive effect of level of development on the creditworthiness of the country. ● The high inflation level adversely affects the rating of the country ● The degree of the economic freedom in the country, thus the institutional strength of the country has a significant positive contribution to the creditworthiness of the country.

Table 4. Estimation of the Debt Intolerance Equation

Model Specification Model I Model 2

Inflation Dummy -3.647*** -2.902**

[-0.9660] [-1.2090]

Debt-to-GDP Ratio -0.224*** -0.245***

[-0.0297] -0.0430]

Debt-to-GDP Ratio Squared 0.000618*** 0.000625**

[-0.0001] [-0.0002]

GDP per Capita 0.000680*** 0.000719**

[-0.0002] [-0.0004]

Default Average Lagged -3.770** -

[-1.5210]

IIR Lagged 0.508*** 0.467***

[-0.0585] [-0.0810]

Time Period 2 5.179** -

[-2.0220]

Time Period 3 3.870** -

[-1.5970]

Time Period 4 4.060*** 0.3

[-1.0690] [-1.2170]

Time Period 5 2.988*** -0.85

[-0.9350] [-1.5210]

Time Period 6 -3.421

34

[-2.0820]

Index of Economic Freedom - 0.204***

[-0.0700]

Constant 25.55*** 18.42**

[-6.0020] [-8.9020]

Observations 330 264

R-squared 0.973 0.973

Robust standard errors in brackets

Statistical Significance levels: *** 99 %, ** 95 %, * 90% According to these models, in addition to the debt to GDP levels, the previous ratings, level of economic development, inflation, default history, and institutional strength are key determinants of the probability of default, and thus, the current sovereign ratings. The models suggest that the contribution of each variable to the ratings differs significantly across countries. Moreover, the negative impact of the debt levels is compensated by other factors which makes some countries more tolerant to the higher levels of debt. On the other hand, the negative impact of the debt levels is scaled up by other factors which makes some countries more debt intolerant. For instance, the significant negative contribution of the high debt-to-GDP ratio to the ratings of the Japan is compensated by the high income level, macroeconomic stability, previous period’s ratings, historical performance in fulfilling its debt obligations as well as strong institution. On the other hand, the negative contribution of the debt-to-GDP ratio to the ratings of Mexico couldn’t be compensated enough by the other variables. Thus, Mexico demonstrates the debt intolerance, as it appears to have low sovereign ratings despite relatively moderate level of debts. The models also suggest that it is not possible to improve the ratings only by decreasing the debt levels particularly for the countries with low to moderate levels of debt. Based on coefficients estimated in the model 1, Table 5 present some non-investment grade countries that cannot achieve an investment grade even with zero public debt, that is, these countries show the high level of debt intolerance and need to improve in other aspects in order to achieve an investment grade. According to the first model, on average, by decreasing debt to GDP ratio by 10 percentage points, the IIR can be improved 2.2 points; and by increasing GDP per capita by 2,000 USD dollar IIR can be improved 2.2 points, holding all other variables constant. Similarly, according to the second model, on average, by improving ranking in Index of Economic Freedom by 10 points, IIR can be improved by 2.

35

Table 5. Estimated IIR with “0” Public Debt

Country IIR Public Debt Estimated IIR (% GDP) with 0 Public Debt

Algeria 50.85 9.06 52.83

Bolivia 38.2 36.16 45.49

Côte d'Ivoire 29 48.87 38.47

Dominican 38.45 34.94 45.52 Republic

Ecuador 27.5 33.83 34.37

Guatemala 41.3 24.2 46.36

Honduras 27.7 46.81 36.83

Myanmar 19.75 34.34 26.71

Nicaragua 22.6 29.4 28.65

Nigeria 38.7 11.5 41.19

Paraguay 42.25 24.19 47.31

Venezuela 26.2 41.5 34.43

36

1 Carmen M. Reinhart, Kenneth S. Rogoff and Miguel A. Savastano. “Debt Intolerance,” National Bureau of Economic Research, August 2003, p.2. 2“Federal Debt: Total Public Debt as a percentage of GDP,” Federal Reserve Bank of St. Louis, April 18, 2017. https://fred.stlouisfed.org/series/GFDEGDQ188S 3 “Central Government Debt, Total (% of GDP) for Mexico,” Federal Reserve Bank of St. Louis, April 18, 2017. 4 Carmen M. Reinhart and Kenneth S. Rogoff, “Growth in Time of Debt,” American Economic Review, May 2010, p.573. https://scholar.harvard.edu/files/rogoff/files/growth_in_time_debt_aer.pdf 5 Carmen M. Reinhart, Kenneth S. Rogoff and Miguel A. Savastano. “Debt Intolerance,” National Bureau of Economic Research, August 2003, p.3. http://www.nber.org/papers/w9908.pdf 6 Id at p.5. 7 Id at p.4. 8 Id at p.3. 9 Id at p.4. 10 Id at p.3. 11 Id. p.5. 12 Ibid. 13 Id. at p.20. 14 Geoffrey J. Bannister and Luis-Diego Barrot, “A Debt Intolerance Framework Applied to Central America, Panama and the Dominican Republic,” International Monetary Fund, 2011 p.3. https://www.imf.org/external/pubs/ft/wp/2011/wp11220.pdf 15 “Abstract,” Economist, 2009. https://www.economist.com/media/pdf/this-time-is-different-reinhart-e.pdf 16 Id. at p.9. 17 Id. at p.10. 18 Id. at p.15. 19 Id. at p.17. 20 Id. at p.18, for instance: "Like Costa Rica, Panama would target a lowering of the debt ratio from 40 to 32.7 to move into the unambiguously investment grade category." 21 Id. at p.17. 22 Duggar et al. “The Causes of Sovereign Defaults: Ability to Manage Crises Not Merely Determined by Debt Levels,” 2010, Moody’s p. 2. 23 p. 6. 24 p. 6. 25 p. 2. 26 p. 6. 27 Raffaela Giordano and Pietro Tommasino, “The Institutional Determinants of Debt Intolerance,” John Wiley & Sons, 2011, p.15. http://onlinelibrary.wiley.com/doi/10.1002/9781118267073.ch2/summary 28 Ibid. 29 p.18. 30 Id. p.16. 31 Ibid. 32 Ibid. 33 Ibid. 34 Id. p.17. 35 Ibid. 36 Ibid. 37 Ibid. 38 “2017 Index of Economic Freedom,” The Heritage Foundation, April 4, 2017. http://www.heritage.org/index/about 39 Ibid.

37

40 Christine Jenkins and Nacha Cattan, “Mexico Risks Credit-Rating Cut as S&P Cites Sluggish Growth,” Bloomberg, August 23, 2016. https://www.bloomberg.com/news/articles/2016-08-23/mexico-at-risk-of-credit- rating-cut-as-s-p-cites-sluggish-growth; Jaime Reusche and Ann Vaan Praagh, “Rating Action: Moody’s Changes Mexico’s Outlook to Negative from Stable; Affirms A3 Rating,” Moody’s Investor Services, March 31, 2016. https://www.moodys.com/research/Moodys-changes-Mexicos-outlook-to-negative-from-stable-affirms-A3-- PR_344609; Dimitra DeFotis, “Mexico Outlook Still Negative, Fitch Says,” Barron’s, February 21, 2017. http://blogs.barrons.com/emergingmarketsdaily/2017/02/21/mexico-debt-outlook-still-negative-fitch-says/ 41 Ibid. 42“Historical Public Debt Database,” International Monetary Fund, April 14, 2017. http://data.imf.org/?sk=806ED027-520D-497F-9052-63EC199F5E63&sId=1390030341854 43 Index of Economic Freedom, “Rule of Law,” The Heritage Foundation, April 18 2017. http://www.heritage.org/index/book/methodology#rule-of-law 44 Terry Miller and Anthony B. Kim et al., 2017 Index of Economic Freedom, The Heritage Foundation, 2017, 23rd Edition, http://www.heritage.org/index/country/mexico 45 Kenneth W. Dam, “The Judiciary and Economic Development,” John M. Olin Law & Economics Working Paper No. 287, The Law School, The University of Chicago, 2nd Series, https://www.brookings.edu/wp- content/uploads/2016/06/200603dam.pdf 46 Ibid. 47 Duncan Wood, “Fighting Corruption in Mexico,” Foreign Affairs, June 22, 2016, https://www.foreignaffairs.com/articles/mexico/2016-06-22/fighting-corruption-mexico 48 Jaime Reusche and Ann Vaan Praagh, “Rating Action: Moody’s Changes Mexico’s Outlook to Negative from Stable; Affirms A3 Rating,” Moody’s Investor Services, March 31, 2016. https://www.moodys.com/research/Moodys-changes-Mexicos-outlook-to-negative-from-stable-affirms-A3-- PR_344609 49 Jonathan Kandell, “José López Portillo, President When Mexico's Default Set Off Debt Crisis, Dies at 83,” NY Times, February 18, 2004, http://www.nytimes.com/2004/02/18/business/jose-lopez-portillo-president-when-mexico-s-default-set-off-debt- crisis-dies-83.html?_r=0 50 James Boughton, “The Mexico Crisis: No Mountain Too High,” International Monetary Fund, October 1, 2001. https://www.imf.org/external/pubs/ft/history/2001/ch07.pdf 51 Horst Kholer, “Mexico - Marking the Retirement of its Brady Bonds,” International Monetary Fund, June 12, 2003. http://www.imf.org/external/np/speeches/2003/061203a.htm 52 Aldo Musacchio, “Mexico’s Financial Crises of 1994-1995,” Harvard Business School. May 8, 2012. https://dash.harvard.edu/bitstream/handle/1/9056792/12-101.pdf?sequence=1 53 James Boughton, “Tequila Hangover: The Mexican Peso Crisis and its Aftermath,” International Monetary Fund, 2012. https://www.imf.org/external/pubs/ft/history/2012/pdf/c10.pdf 54 “Mexico Repays Last Bond From Crises,” The New York Times, June 12, 2003. http://www.nytimes.com/2003/06/12/business/mexico-repays-last-bond-from-crisis.html 55 “Trade Agreements,” Pro Mexico, April 10, 2017. http://www.promexico.gob.mx/en/mx/tratados-comerciales 56 Eric Martin, “Why Lower Oil Prices Don’t Hurt Mexico as Much as They Used to,” Bloomberg, February 24, 2016. https://www.bloomberg.com/news/articles/2016-02-24/why-lower-oil-prices-don-t-hurt-mexico-as-much-as- they-used-to 57 “Revenue Statistics-OECD Countries,” Organization for Economic Co-Operation and Development, April 10, 2017. https://stats.oecd.org/Index.aspx?DataSetCode=REV 58 Alexander Klemm et al., “Mexico Selected Issues,” International Monetary Fund, November 2016, p.13. https://www.imf.org/external/pubs/ft/scr/2016/cr16360.pdf 59 Carmen M. Reinhart and Kenneth S. Rogoff, “This Time is Different, Eight Centuries of Financial Folly,” Princeton University Press, 2009, p.79. 60 Ibid. 61 Moody’s Investor Service, “Moody's affirms Chile's Aa3 government bond ratings and maintains a stable outlook,” 07/11/16, https://www.moodys.com/research/Moodys-affirms-Chiles-Aa3-government-bond-ratings-and- maintains-a--PR_351746 38

62 Ibid. 63 Ibid. 64 Griffen, Emily; Lizano, Esteban; Moe, Kathy-Ann, Siaba, Marta; Stone, Michelle. “Chile: The Success of Story of a ” http://digitalcommons.iwu.edu/cgi/viewcontent.cgi?article=1009&context=parkplace 65 Latam PM. “LatAm Sovereign Credit Ratings Snapshot” https://latampm.com/2016/02/05/latam-sovereign-credit- ratings-snapshot/ 66 International Monetary Fund, World Economic Outlook Database, October 2016, April 8, 2017. 67 The Conference Board Total Economy Database, “Growth Accounting and Total Factor Productivity, 1995- 2015”, November 18, 2016. https://www.conference-board.org/retrievefile.cfm?filename=TED_2_NOV20161.xlsx&type=subsite 68 International Monetary Fund, 2016 Article IV Consultation, https://www.imf.org/external/pubs/ft/scr/2016/cr16376.pdf 69 Ibid 70 Ibid 71 Ibid. 72 International Monetary Fund, 2016 Article IV Consultation, https://www.imf.org/external/pubs/ft/scr/2016/cr16376.pdf 73 Samar Maziad and Atsi Sheth, “Rating Action: Moody’s Changes outlook on Brazil’s Ba2 issuer rating to stable from negative” Moody’s Investor Services, March 15, 2017. 74Anna Snyder, Moody’s Analyst for Brazil. Conference Call. March 23, 2017. 75 2017 Index of Economic Freedom, “Brazil, Quick Facts,” The Heritage Foundation, April 21, 2017 http://www.heritage.org/index/country/brazil 76 IMF Country Report No. 16/348 - 2016 Article IV Consultation. November, 2016. 77 Anna Snyder. Conference Call. March 23, 2017. 78 Fernando J. Cardim de Carvalho, Looking Into The Abyss? Brazil at the mid-2010s, Journal of Post Keynesian Economics, 2016 79 Guilherme Binato, Villela Pedras, “History of public debt in Brazil: 1964 to the present,” Public Debt: The Brazilian Experience, Tesuro Nacional, August 2010 80 Fernando J. Cardim de Carvalho, Looking Into The Abyss? Brazil at the mid-2010s, Journal of Post Keynesian Economics, op.cit. 81 Economist Intelligence Unit, Brazil Country Report, 2017 82 “Research Update: United States of America Long-Term Rating Lowered to ‘AA+’ on Political Risks and Rising Debt Burden; Outlook Negative ,” Standard and Poor’s, August 6, 2011 http://www.standardandpoors.com/en_AP/web/guest/article/-/view/sourceId/6802837 83 Jacob J. Lew. “Managing Our Public Debt Responsibly: A Better Way Forward.” Harvard Journal on Legislation, 2016, p.1. 84 The Economist Online, “Who’s Bigger,” The Economist, June 14, 2012. http://www.economist.com/blogs/graphicdetail/2012/06/daily-chart-8 85 “What are the Sources of Revenue for the Federal Government.” Tax Policy Centre, April 21, 2017. http://www.taxpolicycenter.org/briefing-book/what-are-sources-revenue-federal-government-0 86 Ibid. 87 Carmen Reinhart and Kenneth Rogoff, “Growth in a Time of Debt,” American Economic Review, May 2010, p. 574. https://scholar.harvard.edu/files/rogoff/files/growth_in_time_debt_aer.pdf 88 Ibid. 89 Ibid. 90 2017 Index of Economic Freedom, “United States, Quick Facts,” The Heritage Foundation, April 21, 2017. http://www.heritage.org/index/country/unitedstates 91 Daron Acemoglu and James Robinson, “Why Nations Fail FBBVA Lecture,” MIT, May 21, 2012. https://economics.mit.edu/files/7850 92 Ibid. 93 Jacob J. Lew. “Managing Our Public Debt Responsibly: A Better Way Forward.” Harvard Journal on 39

Legislation, 2016, p.1.

40